发展负责任的生成式人工智能共识

发展负责任的生成式人工智能共识Consensus on Developing ResponsibleGenerative Artificial Intelligence I.总则General 发展负责任的生成式人工智能应始终致力于增进人类福祉，坚持以人为本，推动人类经济、社会和生态可持续发展。应正确认识生成式人工智能所蕴含的巨大潜力和可能风险，遵循统筹发展和安全、平衡创新与伦理、均衡效益与风险的理念，推动生成式人工智能负责任的发展。一方面，应积极推动创新、可持续、包容开放的发展，提升生成式人工智能算力高效、数据高质、算法创新、人才多元、生态开放的能力；另一方面，以高度负责任的态度发展可靠可控、透明可释、数据保护、多元包容、明确责任、价值对齐的生成式人工智能。01 Developing responsible Generative Artificial Intelligence(GenAI) should consistently commit toenhancing human well-being, adhering to a people-centered approach, and promoting the sustainabledevelopment of the economy, society, and ecological environment. I t is important to accurately recognize theimmense potential and possible risks of GenAI. By adhering to a holistic approach that harmonizes developmentwith safety and security, balances innovation with ethical considerations, and fairly evaluates benefits and risks, wecan promote the responsible development of GenAI. On the one hand, innovation-driven, sustainable, inclusive, andopen development of GenAI should be promoted, and the efficient computing power, high-quality data, innovativealgorithm, diverse talents, and an open ecosystem associated with GenAI should be enhanced. On the other hand,GenAI characterized by reliability and controllability, transparency and explainability, data protection, diversity andinclusiveness, accountability, and value alignment should be developed with a highly responsible attitude.01 II.促进生成式人工智能发展Promote the Development of GenAI 积极倡导并稳妥推进生成式人工智能的可持续发展。一是保证经济可持续性。应确保生成式人工智能提高生产力和创造就业机会，提高资源使用效率，实现数实融合的循环经济，推动科技创新，促使经济结构向更高附加值的转变；二是保证社会可持续性。应确保生成式人工智能公平与平等的使用，实现全社会对其的共享共治；三是保证环境的可持续性。生成式人工智能的发展应实现对于自然资源的可持续管理和使用，鼓励采用绿色能源驱动基础设施、提高能源转化效率、绿色开发算法模型应用，降低温室气体排放，实现绿色发展。0202 Actively advocate and prudently promote the sustainable development of GenAI. Firstly, ensureeconomic sustainability. It is vital to ensure that GenAI can enhance productivity and create employmentopportunities, improve the efficiency of resource utilization, achieve a circular economy that integrates digitaltechnologies with the real economy, propel scientific and technological innovation, and facilitate the transfor-mation to an economic structure with higher value added.Secondly, ensure social sustainability. It is crucialto ensure the fair and equitable use of GenAI and to promote co-construction and co-governance at the societallevel.Thirdly, ensure environmental sustainability. The development of GenAI should promote the sustain-able management and use of natural resources, encourage the use of green energy-driven infrastructure, andenhance green R&D of models and applications, to reduce greenhouse gas emissions and achieve green devel-opment. 03构建有益于生成式人工智能健康有序发展的良好环境。一是建立和完善相关的伦理原则和法律法规，重点审视知识产权法律制度，探索人工智能生成物的权利归属方案，对其进行恰当的管理和保护。二是构建包容、扶持、前瞻、可预期的政策环境。为前沿应用孵化构建一个包容的创新环境，为规模推广营造一个优良的营商环境，为赋能经济社会发展搭建一个稳健的监管环境。三是加强国际交流与合作。生成式人工智能的发展需要全球各利益相关方秉持共商共建共享理念，以开放协作态度和举措，开展跨国家、跨领域、跨文化交流与协作，推动形成具有广泛共识的国际评测及标准体系，确保各国共享生成式人工智能的技术惠益。 03Create an environment that is conducive to healthy and orderly development of GenAI. Firstly,establish and optimize relevant ethical principles, laws, and regulations. Legal assessment of the adequacyof the current intellectual property laws should be done with concerned stakeholders to ensure proper attribu-tion of rights to AI-generated objects and to appropriately manage and protect them.Secondly, establish aninclusive, supportive, forward-looking, and predictable policy environment. GenAI requires an inclusiveinnovation environment for incubating its cutting-edge applications, an excellent business environment fordeploying them at scale, and a robust regulatory environment to drive economic and social development.Thirdly, strengthen international communication and cooperation. The development of GenAI requires allglobal stakeholders to uphold the principle of consultation, contribution, and shared benefits, to engage intransnational, cross-disciplinary, and cross-cultural communication and cooperation with open and collabora-tive attitudes and actions, to jointly establish an international assessment and standard system with broadconsensus, to ensure that all countries can share in the technological benefits of GenAI. 04提升生成式人工智能研发及规模应用的能力。一是构建开放共享、普惠包容的算力资源。应推动算力的合理分配与高效利用，降低科技创新的门槛，确保不同地区、不同规模的企业及个人都能获得必要的计算资源。二是推动负责任的数据共享。应鼓励推广高质量数据的共享流动，增强公共数据资源供给，保障数据安全共享与合规利用，提升各领域数据治理水平。三是完善算法创新的设施条件。应前瞻谋划、统筹布局各类平台和开放共享服务网络建设，鼓励算法和基础模型在安全的基础上开源开放，加强跨行业、跨领域协作，推动产学研结合，形成算法创新的良性生态。四是全面加强人才能力建设。针对从业者，应建立人才交流平台，促进互学互鉴与知识共享，设计并实施涵盖多层次、多领域的教育培训项目，增进不同领域技术供需双方的交流与学习。针对公众，应加强科普、教育及培训，提供准确认知，提升数字素养，促进生成式人工智能的普遍接入。五是推动重点领域应用赋能。推动生成式人工智能与各行业数字化场景深度融合，实现应用迭代创新，促进生成式人工智能技术成果在重点领域的应用赋能。 04Enhance the capabilities of R&D and scaling the application of GenAI. Firstly, build open, shared,and inclusively computing resources that are beneficial to all. Efforts should be made to promote the ratio-nal allocation and efficient use of computing power, reducing the barriers to scientific and technological inno-vation. This w